Insights provided by TCAD modelling into on- and off-state performance of pseudo-vertical diamond Schottky barrier diodes
Paper details
Section titled “Paper details”| Field | Value |
|---|---|
| Journal | Diamond and Related Materials |
| Published | 2026-10-06 |
| Authors | G. Verzellesi, Giacomo Biasin, Julien Bassaler, Matteo Buffolo, Vishwajeet Maurya |
| DOI | 10.1016/j.diamond.2026.114274 |
SBD-2 increased room-temperature on-state current density by a factor of 160 at −10 V. These diodes failed at 430 V under reverse bias, compared with 115 V for SBD-1. TCAD analysis suggests that reduced deep-donor compensation explains the higher current. The model also links the higher reverse voltage to fewer acceptor-like surface traps.
Measured gains across two generations
Section titled “Measured gains across two generations”The two generations used different substrates, growth times, and active-layer doping. The reported ranges cover 28 samples with circular and rectangular mesas of different areas.
| Measurement | Condition | SBD-1 | SBD-2 |
|---|---|---|---|
| On-state current density | −10 V, 300 K | (2.9 ± 0.4) A/cm² | (4.6 ± 0.9) × 10² A/cm² |
| On-state current density | −5 V, 300 K | (1.5 ± 0.2) A/cm² | (18 ± 4) A/cm² |
| Rectification ratio | ±10 V, 300 K | > 2 × 10⁴ | > 10¹⁰ |
| Specific on-state resistance | 300 K | (3.7 ± 0.7) Ω cm² | (10 ± 7) × 10⁻³ Ω cm² |
| Schottky barrier height | Temperature-dependent I-V | (0.91 ± 0.09) eV | (1.37 ± 0.02) eV |
| Ideality factor | Temperature-dependent I-V | 1.7 ± 0.2 | 1.8 ± 0.1 |
| Reverse voltage | J = 10⁻⁵ A/cm², 300 K | (16 ± 8) V | (1.6 ± 0.5) × 10² V |
| Device failure voltage | Reverse bias | 115 V | 430 V |
The rectification ratio increased by more than five orders of magnitude. The SBD-2 reverse current at 10 V was dominated by the instrument noise level. The Schottky barrier increased by 0.46 eV, while the ideality factor changed from 1.7 ± 0.2 to 1.8 ± 0.1. The ideality factors and Richardson constants matched literature values for nonannealed contacts, leaving scope for contact improvement.
Device structure and characterization
Section titled “Device structure and characterization”Pseudo-vertical SBDs use a lateral layout and vertical current flow. They avoid conductive substrates required for vertical SBDs and relax the blocking-voltage and current trade-off of lateral SBDs.
SBD-1 used an Ib high-pressure high-temperature substrate from Sumitomo. SBD-2 used an IIa chemical vapor deposition substrate from EDP. Both active layers were grown by microwave plasma chemical vapor deposition. The substrate change aimed to improve crystal quality and benchmark cost-effective CVD substrates for scaling.
Each device had a 0.5 μm heavily boron-doped layer with a boron concentration above 3 × 10²⁰ cm⁻³. SBD-1 used a 2.86 μm drift layer targeted at 3 × 10¹⁶ cm⁻³. SBD-2 used a 2.0 μm drift layer targeted at 5 × 10¹⁵ cm⁻³. Microscopy showed fewer crystallites in SBD-2, suggesting better Schottky contact quality.
Reactive ion etching exposed the buried doped layer. Titanium ohmic contacts were annealed at 600 °C to promote TiC formation and reduce contact resistance. The ohmic stack was Ti/Pt/Au. Ozone treatment provided oxygen termination before Mo Schottky deposition. The Schottky stack was Mo/Pt/Au, and no oxide passivation was used.
A Keysight B1505 measured I-V and C-V characteristics. Ramped mode kept instrument noise below 10 pA. The chuck’s estimated 10 K/W thermal resistance made self-heating negligible. Barrier height came from temperature-dependent I-V fits with the Shockley equation and Richardson-Dushman relation, with series resistance included. C-V measurements covered 0 to 20 V reverse bias at 300 K, 1 MHz, and 50 mV. The SBD-2 depletion region nearly reached the heavily doped layer at 20 V.
TCAD model
Section titled “TCAD model”Two-dimensional Sentaurus Device simulations from the Synopsys TCAD suite solved Poisson and carrier continuity equations with drift-diffusion transport. Cylindrical coordinates made the circular 140 μm SBD-1 and 100 μm SBD-2 devices equivalent to 3D structures.
Quasistationary voltage sweeps included literature models for bandgap, electron affinity, incomplete boron and nitrogen ionization, mobility, and impact ionization. The Schottky boundary included thermionic emission and nonlocal hole tunneling based on the WKB approximation. Deep levels used Shockley-Read-Hall capture and emission.
The incomplete-ionization model was static. Dynamic dopant effects during fast voltage sweeps were not captured by this model. The authors state that a dynamic trap model and transient simulation would be required for that description.
On-state transport
Section titled “On-state transport”Forward-bias measurements and 320 K simulations separated drift-layer doping from deep-donor compensation. Without compensation, nominal SBD-2 had slightly lower on-state current than SBD-1. Measured SBD-2 current was about 12 times higher than SBD-1 at −5 V.
To reproduce thermionic emission, the fitted Mo work function was 5.0 eV for SBD-1 and 4.4 eV for SBD-2. The 0.6 eV model change was comparable with the measured 0.46 eV barrier increase. Fitted hole tunneling mass parameters were 0.6 and 0.5. Deep donors placed 1.7 eV from the conduction band modeled nitrogen impurities. Their concentrations were 2 × 10¹⁶ cm⁻³ in SBD-1 and 2 × 10¹⁵ cm⁻³ in SBD-2. These values produced effective acceptor-minus-donor concentrations of 1 × 10¹⁶ cm⁻³ and 3 × 10¹⁵ cm⁻³.
SBD-2 reached its measured current without reducing the literature maximum room-temperature hole mobility of 2016 cm²/(V s). SBD-1 required a reduced maximum mobility of 1100 cm²/(V s). At high forward bias, measured log-current versus log-voltage slopes approached 1 for SBD-1 and 2 for SBD-2. The authors identify ohmic current in SBD-1 and space-charge-limited current in SBD-2 because of space-charge effects in the p-type drift layer.
Reverse leakage and breakdown
Section titled “Reverse leakage and breakdown”Repeated reverse tests showed degradation in SBD-1 above 100 V. Its first sweep to 115 V increased low-voltage reverse current by a factor of 10⁴. The authors associated this behavior with percolative paths linked to dislocations in the vertical p-type layer. SBD-2 remained stable during repeated sweeps to 400 V, then failed near 430 V. Some failed devices had a burned point at the Schottky contact edge.
TCAD attributed pre-breakdown reverse leakage to hole tunneling from the Schottky contact. Impact ionization became relevant as the electric field approached the avalanche regime. The reverse current depended strongly on the field peak at the metal corner on the exposed diamond surface. Donor-like traps extended the depletion region laterally, reduced the field peak, and shifted breakdown toward higher voltage. Acceptor-like traps pinched the depletion region, increased the field peak, and reduced breakdown voltage.
The surface model used amphoteric dangling-bond traps with 1 eV Gaussian standard deviations. SBD-1 had assumed acceptor and donor peak densities of 1 × 10¹⁴ and 3 × 10¹² cm⁻² eV⁻¹. SBD-2 had acceptor and donor peaks of 1 × 10¹² and 3 × 10¹² cm⁻² eV⁻¹. Thus, SBD-2 required a 100-fold reduction in surface acceptors in the model. Acceptor and donor peaks were 2.5 eV from the conduction band and 1.5 eV from the valence band, respectively.
At 50 V, SBD-1 depletion developed vertically beneath the Schottky contact, while SBD-2 depletion also extended beneath the uncontacted surface. The surface field peak remained 4 to 5 × 10⁶ V/cm in both cases. SBD-2 reached that peak at 300 V, six times the 50 V condition used for SBD-1.
Limits and engineering implications
Section titled “Limits and engineering implications”The SBD-1 reverse-voltage result is reported inconsistently. Table I and the conclusion give (16 ± 8) V, while Section 3 gives (18 ± 4) V for the same current density and temperature. The authors also use uniform average dopant profiles, although C-V showed nonuniform effective doping. These simplifications can affect compensation estimates, depletion width, and field profiles. Nitrogen concentrations may also be overestimated because some boron was inactive.
The surface-trap model is a proof of concept because detailed experimental surface-state data were unavailable. Simulations stopped converging before avalanche onset, so they did not directly predict the final breakdown mechanism. The devices had no dedicated edge termination and no oxide passivation.
SBD-2 had lower targeted boron but higher measured current, so nominal boron concentration alone cannot select a drift layer. Effective doping, donor compensation, and hole mobility must be checked together. For reverse operation, reducing or passivating acceptor-like surface defects and controlling the Schottky edge field are central targets. C-V profiling can reveal reduced effective doping, and repeated high-voltage testing can expose early degradation.
Materials and support
Section titled “Materials and support”For pseudo-vertical diamond Schottky barrier diodes, the required material family is MPCVD-grown boron-doped diamond with a heavily doped contact layer and a lightly doped drift layer. The 6CCVD diamond materials catalog lists boron-doped MPCVD material and metallization with Ti, Au, and Pt, which matches the paper’s boron-doped diamond and Ti/Pt/Au ohmic-contact needs. The paper’s Mo/Pt/Au Schottky stack is outside the listed metallization choices, so that metal and its process require a separate compatibility discussion through engineering support.
View Original Abstract
Pseudo-vertical Schottky barrier diodes were fabricated on boron-doped diamond layers grown by microwave plasma chemical vapor deposition on high-pressure high-temperature and chemical vapor deposition substrates. Analysis of these devices by means of a coupled experimental/simulation approach has allowed on- and off-state operational regimes to be linked to technology-specific parameters, with the aim of providing useful indications for viable optimization routes. Specifically, compensation of boron doping by unintentional donor impurities like nitrogen is suggested to limit the diode on-state conductance, while acceptor surface defects are suggested to critically influence the reverse leakage current and breakdown, so that their reduction or passivation can enable significant breakdown voltage increase.
Boron Doped Diamond On Titanium Bdd Electrode
View Specs →
Boron Doped Single Crystal Diamond Plate
View Specs →
Advanced Thermal Simulation Service Diamond Heat Spreader Performance Analysis
View Specs →References we cover
Section titled “References we cover”- Diamond p-Type Lateral Schottky Barrier Diodes With High Breakdown Voltage (4612 V at 0.01 mA/Mm)
- Design of a normally-off diamond JFET for high power integrated applications